Thai Journal of Operations Research : TJOR

Joint Districting and Transportation Lot-Sizing for A Sugarcane Management System using A Simulated Annealing Algorithm

2024-01-08T09:55:55+07:00

In this paper, we develop the model of the delivery management to transport sugarcane from the cane fields to sugar mill based on the right quantity and right demand for each period. The objective is to minimize the total expected cost. The assumptions of the model are to operate planning in cultivation, harvesting, and transportation. We formulated the decision model of districting and transportation lot sizing problems. In cultivation planning, we propose partitioning the entire sugarcane region by assigning the cane fields and the truck stations into two districts, an inside area, or an outside area. In harvesting, and transportation, we also make decisions of selecting the cane fields, truck stations used to transport, lot sizing during harvesting and transportation under the uncertainty of sugarcane yields for each period. In methodology, we propose the combined simulated annealing algorithm and heuristic (CSAH) to obtain near-optimal solutions. We compare the efficiency of our model with separated models (SM) of districting and transportation lot sizing. The numerical results show the benefits of our model over the separate model based on average total expected cost with 11.45%, savings amounting to 2,826,598 baht.

Enhancing Inventory Management with Multi-Criteria Raw Material Classification: Case Study Approach

2024-04-30T09:30:09+07:00

This research aims to increase the efficiency of raw material storage and reduce the cost of raw material inventory management in the case study company. A multi-criteria inventory classification method based on ABC Analysis combined with FSN Analysis is applied to help classify raw materials. The next step is to analyze the appropriate safety stock to meet the uncertain demand. Techniques for order planning using the min-max system, periodic review system and two bin system was used for the most, moderately, and least important raw material groups, respectively. Comparing the efficiency in terms of total cost of inventory control, it was found that the percentage improvement was equal to 8.70, or equivalent to a total cost of inventory control that was reduced by 61,206 baht. In addition, the proposed method reduces the average raw material inventory level by 38.07 meters or 11.83 percent, resulting in the sunk costs of the case study company being reduced by up to 13,301.56 baht per year.

Prediction of Influenza-Like Illness in Thailand

2024-04-11T14:04:45+07:00

This research aims to analyze the outbreak of influenza in Thailand by studying factors related to the epidemic and predicting the Influenza-like Illness percentage (ILI%). The ILI% data, aggregated monthly for each province in Thailand, is compared using five prediction methods: multiple linear regression, regression tree, Light Gradient Boosting Machine (LightGBM), Extreme Gradient Boosting (XGBoost), and random forest. The features used include vaccine-related factors, risk group disease factors, population factors, and weather factors. Additionally, feature selection methods such as Stepwise, Features Importance Ranking, SHAP Ranking, Boruta, BorutaSHAP, and Mutual Information Scores (MI Scores) with Boruta and BorutaSHAP. To evaluate the performance of the models, the researchers used the symmetric mean percentage error (SMAPE) as a metric. Random forest method, using MI Scores with BorutaSHAP, achieved the lowest SMAPE of 59.11% on the test dataset and identified significant features such as vaccination rate, number of houses, population aged 7-9 years, population aged 15-24 years, and number of patients with stroke. These forecasts can help prevent and mitigate the impact of outbreaks and inform vaccine distribution decisions, as well as community-level outbreak prevention strategies.

Establish a policy for ordering multiple consumable products simultaneously

2024-04-11T15:15:35+07:00

Defining or planning product purchasing policies is very important because setting an appropriate purchasing policy can reduce the company's overall costs. which determines the product ordering policy who will plan or set the policy must know the details of the product and conditions for ordering products with the originating company or manufacturer Including all expenses that must occur in ordering products into the company and must know each product customer behavior to be used to predict future customer orders. This research aims to analyze and establish a policy for purchasing new supplies to be suitable for needs and have the lowest average annual total cost. However, an unstable demand for consumables or Liner bags has zero demand for many periods (Intermittent Demand and Lumpy Demand) cause it difficult for the company to manage inventory and reserve products to prevent product shortages. Therefore, it sets up a policy for ordering new supplies by forecasting demand that may occur in the future. In this research, 4 forecasting methods will be used 1. Forecasting by Croston's method 2. Forecasting by Syntetos and Boylan (SBA) estimation method 3. Forecasting by Shale Boylan and Johnston method (SBJ) 4. Forecasting by the method of Teunter Syntetos and Babai (TSB) and using the mean and standard deviation obtained from the forecasting method with the square root of the mean square error (RMSE) that lowest RMSE is used to calculate the order quantity, reorder point and safety stock to prevent shortages of each item that keeps the total annual cost to a minimum. By using the method of ordering many products at once (Joint Replenishment Problem) under the limitations of a container that can hold only 12 pallets and the ordering requirements that must order a full pallet per item.

Solving Tourist Route Planning in Safari World with Hungarian and Greedy Algorithm

2024-04-11T15:30:55+07:00

The objective of this research is to find the shortest route for touring animal exhibits and attending all shows at Safari World. This problem is addressed by applying the assignment problem and a greedy algorithm. Initially, the assignment problem is utilized to determine the sequence of show visits. Following that, the greedy algorithm is employed to find the optimal route for visiting animal exhibits, maximizing the coverage between the routes to the shows. After executing these two steps, the research produces the shortest route, with a total distance of 7.578 kilometers, and tour starts at 9:00 AM and ends at 4:48 PM.

Forecasting the Supply Chain Demand of Agricultural Residues for Biomass Power Plants Using Machine Learning

2024-04-11T15:51:51+07:00

This research investigated the supply chain of biomass derived from agricultural residues of five crop types in the Northern region of 17 provinces in Thailand, with the aim of evaluating the potential quantity of agricultural residues available for electricity generation through biomass. Historical data was utilized, and 5 machine learning models were employed to determine the optimal model for the dataset, which included 11 factors. The dataset was divided into 80% for training the models and 20% for testing. The Random Forest Regression model with n_neighbors = 1 emerged as the most effective for predicting production from this dataset, achieving a Mean Absolute Percentage Error (MAPE) of 27.511%. The results indicate that machine learning techniques can be utilized to forecast agricultural production effectively. This approach enables the estimation of agricultural residues from various crops, thus facilitating the calculation of biomass energy derived from agricultural residues for biomass power generation. The implications of these findings are substantial, as they demonstrate the potential of machine learning in the utilization of biomass resources for renewable energy production.

Cost-Effectiveness Analysis in Motorcycle Investment for Use in Public Passenger Transport Services

2024-04-11T16:16:55+07:00

Motorcycle taxi service providers face challenges due to the fluctuation of fuel prices, which impact their operations. Additionally, the use of fuel-based motorcycles serves as a catalyst for social and environmental issues. This research employs mathematical principles along with economic analysis to study optimal routes and assess the cost-effectiveness of choosing between fuel-based motorcycles and electric motorcycles with swappable batteries for providing public motorcycle taxi services within the Thammasat University, Rangsit Campus. The objective of this study is to identify suitable routes and analyze the cost-effectiveness of choosing between fuel-based motorcycles and electric motorcycles with swappable batteries for public motorcycle taxi services within Thammasat University, Rangsit Campus. The evaluation of the total cost of ownership for both types of motorcycles over a 10-year lifespan includes operational costs and the external costs associated with carbon dioxide emissions. Additionally, the study considers investment decision criteria by evaluating the payback period, internal rate of return, net present value, and sensitivity analysis throughout the motorcycles' service life in providing motorcycle taxi services. The study reveals that using motorcycles for the transportation of passengers along optimal routes can reduce fuel costs by up to 12.58% annually. Meanwhile, electric motorcycles with swappable batteries have a longer payback period compared to fuel-based motorcycles but offer long-term profitability due to lower total ownership costs and expected future returns, estimated at over 34,089.45 baht. This essential information assists entrepreneurs in making strategic investment decisions regarding the future of motorcycle taxi operations.